Angular Acceleration Sensor Fault Diagnosis Based on LM-BP Neural Network

In practical applications, angular accelerometers may have various failures. It is very important to be able to diagnose these faults in time. BP neural network is widely used in fault diagnosis, however, it has some limitations in angular accelerometer fault diagnosis, such as poor rate of convergence and getting stuck in local minimum. Therefore, a fault diagnosis method based on Levenberg-Marquardt back propagation(LM-BP) neural network is proposed in this paper. By using wavelet packet decomposition and statistical analysis, effective fault diagnosis parameters are determined. In order to verify the effectiveness of the characteristic parameters and the fault diagnosis ability of the LM-BP neural network, six kinds of typical faults of the angular acceleration sensor and its control platform are simulated and tested. The result of experiment shows that this method can validly diagnose angular accelerometer's faults.